KR101383905B1 - method and apparatus for processing server load balancing with the result of hash function - Google Patents

Abstract

In order to load balance a plurality of servers, the server load balancer selects a hash source for a uniform resource locator (URL) obtained from a client, and applies a predetermined hash function to the selected hash source. Then, one of the plurality of servers is selected according to the hash value based on the hash result obtained by applying the hash function, and the client is connected to the selected server.

Description

Method and apparatus for processing server load balancing with the result of hash function}

The present invention relates to a method and apparatus for distributing and processing a server load.

Server load balancing is to distribute the load of the server by efficiently distributing the Internet traffic to a plurality of servers providing the same service, which can increase the server utilization and network bandwidth efficiency.

1 is a diagram illustrating a conventional server load balancing process.

Looking at the server load balancing at layer-4 (L4), as shown in (A) of Figure 1, the load balancer is the IP (Internet protocol) and port (port) information, which is the information of the L4 layer in the incoming packet Refer to to determine which host to send the packet to.

In addition, referring to server load balancing in Layer-7 (L7), as shown in (B) of FIG. 1, the load balancer includes information of the L7 layer in the incoming packet, that is, payload information (for example, payload information in the packet). HTTP header information) to determine which host to send the packet to.

As such, the L7 layer performs server load balancing based on the domain, URL (uniform resource locator) path, file extension, etc. of the HTTP protocol, which is the payload information of the input packet, while storing the content separately for each server. Network address translation (NAT) is widely used to perform server load balancing based on IP and port information.

Recently, the capacity of content through the network has increased and its amount has also increased, and there is a limit on content management with a direct attached storage (DAS) structure in which the entire content must be stored for each server. Accordingly, the content is kept only in the original server, and the service servers use the method of requesting the original server and providing it to the client only when needed for the service. The connection between the edge server and storage has been used such as network attached storage (NAS), storage area network (SAN), HTTP, and cache file system.

However, as the number and capacity of content increases, hitting content decreases in the edge server's local disks and memory, and the load on the edge server increases as the frequency of requesting the content from the origin server increases. In addition, as the number of requests for content from the edge server to the origin server increases, the origin server loads increase.

The problem to be solved by the present invention is to provide a method and apparatus capable of more efficiently balancing the load of the server.

In accordance with an aspect of the present invention, a server load balancing processing method includes: receiving, by a server load balancer, a uniform resource locator (URL) from a client; The server load balancer selecting a hash source from the URL and applying a preset hash function to the selected hash source; Obtaining, by the server load balancer, a hash value obtained by dividing a hash result obtained by applying the hash function by the total number of servers as a hash value; Selecting, by the server load balancer, one of a plurality of servers in response to the hash value; And the server load balancer connecting the client to the selected server.

In addition, the server load balancing processing method according to another aspect of the present invention, the server load balancer receiving a uniform resource locator (URL) from the client; The server load balancer selecting a hash source from the URL and applying a preset hash function to the selected hash source; Obtaining, by the server load balancer, a hash value obtained by dividing a hash result obtained by applying the hash function by a first number greater than the total number of servers; Selecting, by the server load balancer, one of a plurality of servers in response to the hash value; And the server load balancer connecting the client to the selected server.

The selecting may include selecting the server corresponding to the hash value from a server matching table in which the remaining obtainable values are divided into a plurality of groups, and one server is matched with each of the remaining value groups. The range of each group dividing the remaining obtainable values from the server mapping table may vary according to the total number of servers.

Server load balancing processing method according to another aspect of the invention, the server load balancer receiving a uniform resource locator (URL) from the client; The server load balancer selecting a hash source from the URL and applying a preset hash function to the selected hash source; Acquiring, by the server load balancer, a hash result obtained by applying the hash function as a hash value; Checking, by the server load balancer, whether there is a server previously selected for the hash value from a history table, based on the hash value; If there is a server previously selected for the hash value, selecting the selected server from a plurality of servers; And the server load balancer connecting the client to the selected server.

Here, when there is no server previously selected for the hash value, the method may further include selecting one server in consideration of a load state of the plurality of servers. On the other hand, the step of selecting a server in consideration of the load condition comprises the steps of checking the available capacity for each server based on the load and outgoing traffic of each of the plurality of servers; And selecting a server having the highest value among the available capacities of the servers.

In server load balancing processing methods having such a feature, the method may further include determining whether the service is available for the selected server, and in this case, connecting the client may be performed only when the selected server is serviceable. You can connect the client to the selected server.

The determining may include determining that the service is not available when the server is in an unserviceable state. The unserviceable state is a system in which a server fails physically and software to perform a service. Disorder status; Server overload condition, in which the server is physically / softwarely operational, but additional services are difficult because the service is already serving the entire service available capacity; Although the server is physically / softwarely operating, it may be at least one of a state in which a service cannot be performed because there is no content to be serviced.

Meanwhile, when the selected server is not serviceable, the method may further include generating a new hash source to which the hash function is to be applied, and applying a hash value obtained by applying a hash function to the newly generated hash source. Server selection can be made based on this.

The generating of the new hash source may generate a hash result obtained by applying the hash function as a new hash source.

The generating of the new hash source may use a hash result obtained by applying the hash function and a string including the URL as a new hash source.

The step of connecting the client may include requesting content corresponding to the parameter while providing a parameter obtained from the URL from the client to the selected server; And receiving a response corresponding to the request from the server and delivering the response to the client.

In accordance with still another aspect of the present invention, an apparatus includes a server load balancer configured to perform load balancing on a plurality of servers, and a data obtainer configured to receive a uniform resource locator (URL) from a client; A function application unit which selects a hash source from the URL and applies a preset hash function to the selected hash source; A server selector configured to select one of the plurality of servers according to a hash value based on a hash result obtained by applying the hash function; And a delivery unit connecting the client to the selected server.

The server selector may use the remaining value obtained by dividing the hash result obtained by applying the hash function by the total number of servers as the hash value to select the server.

The server selecting unit may further include a server matching table that divides the remaining obtainable values into a plurality of groups and matches one server with respect to each of the remaining value groups, and applies a hash function obtained by applying the hash function to the server. The remaining value obtained by dividing by the first number greater than the total number of may be obtained as a hash value, and a server corresponding to the hash value may be selected from the server matching table.

The server selecting unit may further include a history table in which a server selection history including information on a server selected for each hash value is stored, and the hash result obtained by applying the hash function as a hash value. From the table it is possible to select a server that was previously selected for the hash value. In this case, when there is no server previously selected for a hash value in the history table, the server selecting unit checks the available capacity of each server based on load and outgoing traffic for the plurality of servers, and the available capacity of each server. You can choose the server with the highest value among them.

Meanwhile, when the selected server is in an unserviceable state, the function applying unit may use a hash result obtained by applying the hash function as a new hash source and obtain a new hash result by applying the hash function.

The function application unit may use a hash result obtained by applying the hash function and a string including all or part of the URL as a new hash source and apply the hash function when the selected server is in an unserviceable state. The result can be obtained.

The function application unit may use at least one of a protocol, a domain or an IP, a folder, a file name, and a parameter when selecting a hash source from the URL.

In addition, the server selection unit may use the ISP, region, and country information of the client IP for server selection.

In the apparatus according to this aspect of the invention, the selected server may be a server group that physically includes one or more servers.

According to an embodiment of the present invention, even if the number and capacity of the contents increase, the load on the server can be effectively distributed. In addition, server and storage efficiency can be maximized.

1 is a diagram illustrating a conventional server load balancing process.
2 is a diagram illustrating a structure of a server load balancer according to an exemplary embodiment of the present invention.
3 is an exemplary diagram illustrating a server matching relationship based on a hash value.
4 is an exemplary view showing a result of applying a hash function according to an embodiment of the present invention.
5 is a flowchart illustrating a server load balancing processing method according to a first embodiment of the present invention.
6 is a flowchart of a server load balancing processing method according to a second embodiment of the present invention.
7 is a flowchart illustrating a server load balancing processing method according to a third embodiment of the present invention.
8 is a flowchart illustrating a server load balancing processing method according to a fourth embodiment of the present invention.
9 is an exemplary diagram illustrating a server connection process according to an embodiment of the present invention.
10 is an exemplary diagram illustrating another server connection process according to an embodiment of the present invention.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings so that those skilled in the art can easily carry out the present invention. As those skilled in the art would realize, the described embodiments may be modified in various different ways, all without departing from the spirit or scope of the present invention. In order to clearly illustrate the present invention, parts not related to the description are omitted, and similar parts are denoted by like reference characters throughout the specification.

Throughout the specification, when an element is referred to as "comprising ", it means that it can include other elements as well, without excluding other elements unless specifically stated otherwise.

Hereinafter, a server load balancing processing method and apparatus thereof according to an embodiment of the present invention will be described with reference to the drawings.

In an embodiment of the present invention, a server load is distributed by using a hashing method for a hyper text transfer protocol (HTTP) uniform resource locator (URL) input from a client. The hashing method obtains a hash result by applying a hash function to an HTTP URL of an input packet. In an embodiment of the present invention, load balancing is performed on a server based on a hash value according to the hash result.

An HTTP URL represents a protocol for telling where a resource is on an HTTP network. It has the following structure:

HTTP URL = Protocol + domain name + file path + get parameter

The protocol here is “http: //”, where the hostname represents the domain name. The file path includes a folder and a file name, and the folder and file name are separated by “/”. Get parameters start with "?", And each parameter is separated by "&" when representing multiple parameters.

An HTTP URL having such a structure may be expressed as, for example, "http://vod.domain.com/vod/sample.mp4?id=1234", where "http" represents a protocol and "vod. domain.com ”represents a domain name,“ /vod/sample.mp4 ”represents a file path, and“ id = 1234 ”represents a parameter.

In an embodiment of the present invention, a hash function (or a hash algorithm) is applied to a URL having the above structure to obtain a hash value. Hash functions create a kind of short "electronic fingerprint" from arbitrary data, and produce results using methods such as cutting, replacing, or repositioning the data. These results are called "hash results." In an embodiment of the present invention, a hash value for server selection is obtained based on the hash result. The hash function works deterministically, if the two hash results are different, the original data for that hash result must be different, and vice versa. The reliability of a hash function is determined by how many hash collisions, that is, when the two different data have the same hash result. The more hash collisions, the more difficult it is to distinguish different data and the cost of retrieving the data can increase. have.

A hash function basically has a deterministic characteristic that the original input data is different if the two hash results are different, but it is not a singular function. If they have the same hash result, it suggests that the original input data are the same, but does not guarantee. Even if only one bit of the original data is changed, the hash result is very different due to the nature of the hash function.

2 is a diagram illustrating a structure of a server load balancer according to an exemplary embodiment of the present invention.

As shown in FIG. 2, the server load balancer 1 according to an exemplary embodiment of the present invention is connected to a plurality of clients 2 through a network and also to a plurality of servers 3.

The server load balancer 1 obtains a hash result by applying a hash function to data input from the client 2, and sends the client to one server among a plurality of servers based on the hash value obtained according to the hash result. Connect

To this end, the server load balancer 1 is obtained as a data obtaining unit 11 for obtaining data from the client 2, a function applying unit 12 for applying a hash function to the obtained data, and a result of applying the hash function. The server selector 13 selects a server based on the hash value, and the connection unit 14 transmits a client's request to the selected server.

The function applying unit 12 selects a part to which the hash function is applied from the obtained data, and applies a hash function to the selected part to obtain a result. The result is used as a hash value or a value obtained by performing a predetermined process on the result is used as a hash value.

Meanwhile, in order to select a server according to the hash value, a predetermined hash value is matched with respect to the plurality of servers 3. 3 is an exemplary diagram illustrating a server matching relationship based on a hash value. As illustrated in the accompanying FIG. 3, when the hash value is, for example, ①, ②, ③, one server 31 is matched to correspond to the hash value of "①," and the value of "②," One server 32 may be matched to the hash value, and one server 33 may be matched to the hash value of "③". In this state, one server is selected from the plurality of servers 31 to 33 according to the hash value obtained from the data input from the client 2. When the input data is the same URL, since the same hash value is always obtained, the client's request can be delivered to the same server.

In the server load balancer 1, an example of applying a hash function to input data will be described.

4 is an exemplary view showing a result of applying a hash function according to an embodiment of the present invention.

The data extracted from the packet transmitted from the client 2, that is, the input data URL is a moving image as “http://vod.domain.com/vod/sample.mp4?id=1234” as shown in Fig. 4A. Suppose it is related to content. In the network with a plurality of servers, for example N (where N is a positive integer) of servers (server [0] to server [N-1]), the server selected according to the hash function is selected as server [ (int) hash function (HTTP URL)% N].

The hash function is applied to the file path “vod / sample.mp4” from the input data “http://vod.domain.com/vod/sample.mp4?id-1234”. The portion of the input data (URL) that the hash function applies to is called the "hash source". The file path of the URL is used here as the hash source, but is not limited thereto. For example, other URL components such as the domain name of the HTTP URL and the information of the HTTP header such as the user agent may be additionally used as necessary.

When the hash function “MD5” is applied to the file path “vod / sample.mp4” of the input URL, 32 character strings are obtained as shown in Fig. 4B, while using “CRC32” as the hash function. In this case, an integer of 10 digits can be obtained as shown in (C) of Fig. 4. A hash result of “3756904710” is obtained using “CRC32” as a hash function, and the obtained hash result is the number of servers “N”. In this case, a hash value of “3756904710% 4 = 2”, that is, a hash value is obtained, and a server [2] may be selected according to the hash value “2” thus obtained.

In an embodiment of the present invention, a server to which a client's request is delivered is selected based on a hash value obtained by applying a hash function to an input data URL, and the client's request is transmitted to the selected server. The forwarding of the client's request to the server here includes all the related operations of connecting the server and the client, which will be described in more detail later.

Next, a server load balancing processing method according to an embodiment of the present invention will be described.

In this example, it is assumed that there are N servers (or server groups), and that hash values S [0] to S [N-1] are assigned to each server, respectively. It demonstrates.

First, a server load balancing processing method according to a first embodiment of the present invention will be described. 5 is a flowchart illustrating a server load balancing processing method according to a first embodiment of the present invention.

The server load balancer 1 receives input data, that is, an HTTP connection request from the client 2 (S100). Obtain a URL from the HTTP connection request, and obtains a hash source to apply the hash function from the obtained URL (S110). The file path of the URL is used here as the hash source, but is not necessarily limited thereto.

The server load balancer 1 obtains a hash result by applying a hash function preset to a hash source (S120), and obtains a hash value for selecting a server from the obtained hash result.

In the first embodiment of the present invention, the hash value obtained by dividing the hash result obtained by applying the hash function to the hash source by “N” which is the total number of servers is used as the hash value (S130). For example, servers S0, S1, S2, and S3 exist, and the hash source “/ vod / sample obtained from the URL of“ http://vod.domain.com/vod/sample.mp4?id=1234 ”. If the "3756904710" hash result is obtained by applying a hash function (e.g. CRC32) for .mp4 ", the remaining value" 2 "obtained by dividing the hash result" 3756904710 "by" 4 "is used as the hash value. .

The server load balancer 1 selects one server corresponding to the hash value obtained from the plurality of servers (S140). The client request is forwarded to the selected server (S150). For example, the server S2 corresponding to the hash value "2" is selected, and the client is connected to the selected server S2.

In the first embodiment of the present invention, it is possible to efficiently distribute the load of the servers simply.

Next, a server load balancing processing method according to a second embodiment of the present invention will be described. 6 is a flowchart of a server load balancing processing method according to a second embodiment of the present invention.

The second embodiment of the present invention provides a method for easily performing server load balancing even when the number of servers (or server groups) is changed.

As in the first embodiment described above, as shown in FIG. 6, the server load balancer 1 receives an HTTP connection request from the client 2, obtains a hash source from the URL, and obtains the URL from the obtained URL. By applying a hash function to obtain a hash value (S200 ~ S220).

The server load balancer 1 obtains a hash value by applying a hash function preset for the hash source. In the second embodiment of the present invention, a hash result obtained by applying a hash function to a hash source is divided by “M (where M is a positive integer)”, which is a sufficiently large number than “N”, which is the total number of servers. A value of M-1 is used as a hash value (S230). In particular, the second embodiment of the present invention uses the remaining values obtained by dividing the hash result by M which is greater than N, which is the total number of servers, so that the remaining values are larger than the total number of servers. Accordingly, the server load balancer 1 may previously associate the servers with the remaining values obtainable using the server matching table, and select the server from the server matching table based on the hash value obtained through the above step S230. That is, the remaining values that can be obtained are divided into a plurality of groups, a table in which one server is matched with each other of the remaining groups of values is generated in advance, and then a server is selected from the server matching table based on the hash value. For example, if there are S0, S1, S2, and S3 servers, and M = 100, the server matching table shown in Table 1 may be used.

Rest of the value
(Hashed% M) server
(Or server group) 0-24 S0 25-49 S1 50-74 S2 75-99 S3

As shown in Table 1, the remaining values 0 to 100 that can be obtained are divided into four groups (first group (0 to 24), second group (25 to 49), third group (50 to 74), and fourth group ( 75 to 99)), and a server matching table in which one server is matched with each group may be used. The server matching table may be stored and managed in the server selector 13 of the server load balancer 1.

As in the first embodiment, when the hash result of “3756904710” is obtained by applying the hash function to the hash source “/vod/sample.mp4 '”, the hash result “3756904710” is set to “100” of M. The remaining value "10" obtained by dividing is used as the hash value.

The server load balancer 1 selects one server corresponding to the hash value obtained from the plurality of servers by using the server matching table preset as described above (S240). The client request is forwarded to the selected server (S250). For example, the server S0 corresponding to the hash value "10" is selected from the server matching table shown in Table 1, and the client is connected to the selected server S0.

According to the embodiment of the present invention, even if the number of servers is changed, it is possible to efficiently perform server load balancing. In detail, even when the number of servers (or server groups) is changed from N to M, the server load may be effectively distributed by converting the server matching table. For example, the number of servers is N, the content is cached in local storage (for example, disk, memory, etc.) of the N servers, and server load balancing is performed in the same manner as in the first embodiment. If the number of servers is changed to M, there may be a problem of re-cached the content cached in the existing servers, in which case the processing of the client request may be delayed.

However, according to the second embodiment of the present invention, even if the total number of servers is changed, the range of the remaining values matched to the server in the server matching table may be adjusted.

Assume that the number of servers is increased to five while the server matching table shown in Table 1 is maintained and managed. In this case, the server matching table can be changed as shown in Table 2 below.

Rest of the value
(Hashed% M) server
(Or server group) 0-19 S0 25-44 S1 50-69 S2 75-94 S3 20-24, 45-49,
70 ~ 74, 95 ~ 99 S4

Even when the number of servers is reduced to three, the server matching table may be changed as shown in Table 3 below.

Rest of the value
(Hashed% M) server
(Or server group) 0-19, 75-82 S0 25 ~ 44, 83 ~ 90 S1 50 ~ 69, 91 ~ 99 S2

As described above, according to the embodiment of the present invention, the range of the remaining values matched to each server in the server matching table can be adjusted according to the number of servers used, and adaptive server load balancing processing is possible even when the number of servers is changed. Do.

Next, a server load balancing processing method according to a third embodiment of the present invention will be described. 7 is a flowchart illustrating a server load balancing processing method according to a third embodiment of the present invention.

As in the first embodiment described above, as shown in FIG. 7, the server load balancer 1 receives an HTTP connection request from the client 2, obtains a hash source from the URL, and obtains from the obtained URL. Hash values are obtained by applying a hash function (S300 ~ S320).

The server load balancer 1 obtains a hash value by applying a hash function preset for a hash source, selects a server if there is a server previously selected for the obtained hash value, and selects a corresponding server for the hash value. If no server is selected, the server selects a random server considering the load of the server. To this end, the third embodiment of the present invention stores and manages a history table that stores server selection history. The server selection history includes information on servers that have been previously selected for each hash value. For example, the server selector 13 of the server load balancer 1 may store and manage a history table, and the history table may be implemented in a database form.

The history table according to an embodiment of the present invention may be as Table 4 below.

Hash result
server
(Or server group) 1546554652 S0 5456446846 S1 3756904710 S2

As illustrated in Table 4, the history table according to an embodiment of the present invention stores a server that has been selected corresponding to a hash result obtained by applying a hash function to a hash source. Here the hash result is used as the hash value for the server selection.

For example, servers S0, S1, S2, and S3 exist, and the hash source “/ vod / sample obtained from the URL of“ http://vod.domain.com/vod/sample.mp4?id=1234 ”. If a "3756904710" hash result is obtained by applying a hash function (e.g. CRC32) for .mp4 '", check whether there is a server selected from the history table based on the hash result" 3756904710 ". As shown in Table 4, the history table from which the server S4 was selected corresponding to the hash result “3756904710” is checked from the history table. Thereafter, the server S4 is selected as a server to receive the input client request.

Based on this process, the server load balancer 1 uses a hash result obtained by applying a hash function preset to a hash source as a hash value, and checks whether there is a server selected from the history table corresponding to the hash value. Check (S330, S340).

As a result of the check, if there is a server previously selected corresponding to the currently obtained hash value in the history table, the corresponding server is selected (S350). On the other hand, if there is no server previously selected corresponding to the hash value in the history table, the server load balancer 1 selects one server in consideration of the load status of each server (S360).

In this case, the load state of the server represents available capacity considering loads and outgoing traffic of each server. In order to manage the load status of the server, the server load balancer 1 may comprise a server available capacity database. The server available capacity database may be configured as shown in Table 5 below.

server
(Or server group) Usable capacity S0 30 S1 70 S2 30 S3 50

The western load balancer 1 checks the load status of each server from the server available capacity database, and selects the server with the highest available capacity. For example, among the servers of S0, S1, S2, and S3, a server S1 having the highest available capacity may be selected (S370). The server load balancer 1 stores a history of the selected server in a history table. That is, the server selected according to the load state of the server corresponding to the currently obtained hash value is newly stored corresponding to the history table.

Thereafter, the server load balancer 1 transmits a client request to the selected server (S380).

According to the third embodiment of the present invention, by using the existing server selection history according to the client request, it is possible to maximize the content cached in the storage of each server, and to improve the responsiveness to the client request. have.

In addition, even if there is no existing server selection history, by selecting a server based on the available capacity of each server to process a client request, it is possible to minimize the load on a specific server (or server group).

Next, a server load balancing processing method according to a fourth embodiment of the present invention will be described. 8 is a flowchart illustrating a server load balancing processing method according to a fourth embodiment of the present invention.

In the server load balancing processing method according to the fourth embodiment of the present invention, the client may be connected by determining whether the service is available to the selected server without directly connecting the client to the selected server in response to the client request.

According to the above-described embodiments, the selected server may be in a service-unavailable state. Some examples are: First, a server may be in a system failure state that is physically and / or software-failed and unable to perform service. Second, although the server is physically / softwarely operating normally, additional services may be difficult due to already serving the entire service available capacity. Lastly, the server may be physically / softwarely operated, but the service may not be performed because the content to be serviced does not exist.

Therefore, the server load balancer 1 according to the fourth embodiment of the present invention may determine the non-service state as described above, and additionally select another server when the selected server is in the non-service state.

As in the first embodiment described above, as shown in FIG. 8, the server load balancer 1 receives an HTTP connection request from the client 2, obtains a hash source from the URL, and obtains from the obtained URL. Hash values are obtained by applying a hash function (S400 to S420).

The server load balancer 1 obtains a hash value by applying a hash function preset to a hash source (S430), and selects one server based on the obtained hash value (S440). The method of selecting a server may be one of the methods according to the first to third embodiments described above.

The server load balancer 1 determines whether service is possible for the selected server (S450). If the selected server does not correspond to the non-serviceable state as described above, it is determined as a serviceable state and transmits a client request to the selected server (S460 and S470).

On the other hand, if the selected server corresponds to one of the unserviceable states as described above, it is determined that the service is not available, and server selection is performed again. To this end, the server load balancer 1 regenerates a hash source to which the hash function is applied (S480).

In more detail, the previous hash result obtained by applying the hash function in step S420 or a string including the previous hash result is used as a new hash source to which the hash function is applied. And recursively handle hash functions. For example, if the previous hash result is “Hash (URL)”, a hash function of “Hash (Hash (URL))” may be obtained by applying a hash function using the previous hash result as a hash source.

The server is selected based on the hash value according to the newly obtained hash result. In this case, when the newly selected server is also in an unserviceable state, a server selection process of creating a new hash source again and selecting a server based on this may be repeatedly performed until a serviceable server is selected. In this case, for example, the hash source generation process may be repeatedly performed based on the following equation. (number of iterations until n possible servers are selected)

Based on Equation 1,

Primary Hash Result: Hash (URL)

Second hash result: Hash (Hash (URL))

Third hash result: Hash (Hash (Hash (URL)))

...

nth hash result: Hash (Hash (URL, n-1))

A hash result generation in the form of, may be performed, and a server selection process based on the generated hash result may be performed.

Meanwhile, when generating a new hash source, a hash result may be obtained using a string including a previous hash result and a URL obtained from a client as a hash source. For example, a hash source including the previous hash result “Hash (URL)” and “URL” may be applied to the hash function to obtain a hash result of “Hash (Hash (URL) + URL)”. Such a hash source generation can be used when a hash function has a relatively smaller range of hash results than other hash functions, or when a hash value is obtained differently from when a hash function is first applied. It doesn't happen.

This hash source generation process may be repeatedly performed based on, for example, the following equation. (number of iterations until n possible servers are selected)

Based on Equation 2,

Primary Hash Result: Hash (URL)

Second hash result: Hash (Hash (URL) + URL)

Third Hash Result: Hash (Hash (Hash (URL) + URL) + URL)

...

nth hash result: Hash (Hash (URL, n-1) + URL)

The hash source and hash result generation in the form of may be performed repeatedly.

In addition, if the client IP is additionally reflected in the server selection in addition to the hash result for the URL, it is possible to divide the load by ISP, country and region.

Next, in the server load balancing processing method according to an embodiment of the present invention as described above, a process of connecting the client to the server by transmitting the client's request to the selected server will be described in more detail.

9 is an exemplary diagram illustrating a server connection process according to an embodiment of the present invention.

The server load balancer 1 according to an embodiment of the present invention receives the connection request packet input from the client 2 and connects the client 2 to the selected server by changing a destination address of the received packet. Can be.

As illustrated in FIG. 9, the server load balancer 1 is implemented in the form of a built-in L4 load balancing function based on L7 to connect L7 between the client 2 and the server load balancer 1. do. The server load balancer 1 requests the content corresponding to the GET parameter obtained from the connection request HTTP URL input from the client 2 to the selected server (for example, S0) among the plurality of servers, and requests from the server. The response to the request is provided to the client (2).

According to this connection process, the client 2 does not need to know anything about the L7 load balancing structure, and the response through the selection server is provided through the server load balancer 1 having the L7 load balancing structure for the client's HTTP request. . In particular, this connection process is not large in traffic volume, and can be used more effectively when serving low-volume content such as HTML, text, and images.

10 is an exemplary diagram illustrating another server connection process according to an embodiment of the present invention.

The server load balancer 1 according to the embodiment of the present invention receives a connection request packet input from the client 2, and provides a redirect URL including domain or IP information about the server selected by the client 2. The client 2 then connects to the selected server.

This connection process may be performed when the client 2 supports HTTP 301/302 redirection, not a program that merely performs a function of requesting a page to a server according to TCP.

If the client 2 supports HTTP 301/302 redirection, the server load balancer 2 goes to the selected server based on the hash source obtained from the HTTP URL input from the client 2, as illustrated in FIG. In order to connect a client, a redirect URL containing information about the selected server is provided to the client 2. The client 2 connects to the corresponding server according to the redirection URL provided from the server load balancer 2.

This connection process is very large and can be effectively used for transmitting and receiving unstructured data such as high quality images.

The embodiments of the present invention are not limited to the above-described apparatuses and / or methods, but may be implemented through a program for realizing functions corresponding to the configuration of the embodiment of the present invention, a recording medium on which the program is recorded And such an embodiment can be easily implemented by those skilled in the art from the description of the embodiments described above.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, It belongs to the scope of right.

Claims (25)

Receiving, by the server load balancer, a uniform resource locator (URL) from a client;
The server load balancer selecting a hash source from the URL and applying a preset hash function to the selected hash source;
Obtaining, by the server load balancer, a hash value obtained by dividing a hash result obtained by applying the hash function by the total number of servers as a hash value;
Selecting, by the server load balancer, one of a plurality of servers in response to the hash value; And
The server load balancer connecting the client to the selected server
Including, server load balancing processing method. Receiving, by the server load balancer, a uniform resource locator (URL) from a client;
The server load balancer selecting a hash source from the URL and applying a preset hash function to the selected hash source;
Obtaining, by the server load balancer, a hash value obtained by dividing a hash result obtained by applying the hash function by a first number greater than the total number of servers;
Selecting, by the server load balancer, one of a plurality of servers in response to the hash value; And
The server load balancer connecting the client to the selected server
Including, server load balancing processing method. The method according to claim 2, wherein
The step of selecting
And dividing the remaining obtainable values into a plurality of groups and selecting a server corresponding to the hash value from a server matching table in which one server is matched corresponding to each of the remaining value groups. The method of claim 3, wherein
And the range of each group dividing the obtainable remaining values in the server matching table is varied according to the total number of servers. Receiving, by the server load balancer, a uniform resource locator (URL) from a client;
The server load balancer selecting a hash source from the URL and applying a preset hash function to the selected hash source;
Acquiring, by the server load balancer, a hash result obtained by applying the hash function as a hash value;
Checking, by the server load balancer, whether there is a server previously selected for the hash value from a history table, based on the hash value;
If there is a server previously selected for the hash value, selecting the selected server from a plurality of servers; And
The server load balancer connecting the client to the selected server
Including, server load balancing processing method. The method of claim 5, wherein
If there is no server previously selected for the hash value, selecting one server in consideration of the load state of the plurality of servers. The method of claim 6, wherein
Selecting one server in consideration of the load state
Identifying available capacity for each server based on load and outgoing traffic of each of the plurality of servers; And
Selecting a server having the highest value among the available capacities of the servers
Including, server load balancing processing method. The method according to any one of claims 5 to 7,
And a server selection history including information on a server selected for each hash value. The method according to any one of the preceding claims.
Determining whether the service is available for the selected server;
The step of connecting the client connects the client to the selected server only if the selected server is serviceable. The method of claim 9, wherein
The determining may determine that service is not possible when the state of the server is in an unserviceable state.
The out of service state
A system failure state in which a server fails physically and software and cannot perform services;
Server overload condition, in which the server is physically / softwarely operational, but additional services are difficult because the service is already serving the entire service available capacity; And
The server is in normal physical / software operation but cannot be serviced because the content to be serviced does not exist.
Server load balancing processing method. The method of claim 9, wherein
If the selected server is not serviceable, creating a new hash source to apply the hash function to
Further comprising:
And server selection is performed based on a hash value obtained by applying a hash function to the new hash source. The method of claim 11, wherein
Generating the new hash source
And generating a hash result obtained by applying the hash function as a new hash source. The method of claim 11, wherein
Generating the new hash source
And a hash result obtained by applying the hash function and a string including the URL as a new hash source. The method according to any one of the preceding claims.
The step of connecting the client
Requesting content corresponding to the parameter while providing a parameter obtained from the URL from the client to the selected server; And
Receiving a response corresponding to the request from the server and forwarding the response to the client
Including, server load balancing processing method. The method according to any one of the preceding claims.
The step of connecting the client
Providing a redirect URL to the client that includes domain or IP information of the selected server
Including, server load balancing processing method. In a server load balancer that performs load balancing for a plurality of servers,
A data acquisition unit for receiving a uniform resource locator (URL) from a client;
A function application unit which selects a hash source from the URL and applies a preset hash function to the selected hash source;
A server selector configured to select one of the plurality of servers according to a hash value based on a hash result obtained by applying the hash function; And
A transfer unit connecting the client to the selected server
Comprising a server load balancer. The method of claim 16, wherein
And the server selector uses the remaining value obtained by dividing the hash result obtained by applying the hash function by the total number of servers as the hash value to select the server. The method of claim 16, wherein
The server selector
Further comprising a server matching table dividing the remaining obtainable values into a plurality of groups and matching one server corresponding to each of the remaining value groups;
A server load is obtained by dividing a hash result obtained by applying the hash function by a first number greater than the total number of servers as a hash value and selecting a server corresponding to the hash value from the server matching table. Dispersing device. The method of claim 16, wherein
The server selector
The table further includes a history table in which a server selection history including information of a server selected for each hash value is stored.
And a hash result obtained by applying the hash function as a hash value, and selecting a server previously selected for the hash value from the history table. The method of claim 19, wherein
The server selector
If there is no server previously selected for a hash value in the history table, the available capacity for each server is checked for each of the plurality of servers based on load and outgoing traffic, and the highest value among the available capacity for each server is determined. Selecting a server having a server load balancer. The method according to any one of claims 16 to 20.
The function application unit
And when the selected server is in an unserviceable state, uses the hash result obtained by applying the hash function as a new hash source and obtains a new hash result by applying the hash function. The method according to any one of claims 16 to 20.
The function application unit
When the selected server is out of service, using a hash result obtained by applying the hash function and a string including all or part of the URL as a new hash source and applying the hash function to obtain a new hash result, Server Load Balancer. The method according to any one of claims 16 to 20.
The function application unit
And selecting a hash source from the URL, using at least one of a protocol, a domain or an IP, a folder, a file name, and a parameter. The method according to any one of claims 16 to 20.
In the server selector
A server load balancer that uses the ISP, region, and country information of the client's IP for server selection. The method according to any one of claims 16 to 20.
The selected server
A server load balancer, which is a group of servers that physically contains one or more servers.

Images